ABSTRACT Measuring the value of cabinet portfolios is a fundamental task in the study of coalition politics. Yet existing measures are often static, have limited coverage, or assume that all parties value portfolios equally. In this paper, we introduce a new approach that overcomes these problems. We argue that parties signal their portfolio preferences through legislative behaviors, such as floor speeches, bill sponsorship, motions, and budgetary earmarks. Leveraging these behaviors as inputs, we develop and validate a dynamic Bayesian latent‐variable model to disentangle two concepts: a portfolio's underlying salience and the specific value each party assigns to it, independent of the effects of office‐holding. Applying this method to Brazil, a presidential system in which government coalitions and cabinet changes are frequent, we show that our estimates capture portfolio importance and their temporal shifts better than alternative measures. Furthermore, we show that portfolio allocation among coalition parties is more proportional once we weight portfolios using our salience scores. We provide a checklist and a replication repository applied to Denmark for researchers interested in adapting our framework to other contexts.
Meireles et al. (Tue,) studied this question.